[
https://issues.apache.org/jira/browse/QPID-1344?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12638541#action_12638541
]
Rafael H. Schloming commented on QPID-1344:
-------------------------------------------
I'd prefer wrapping rather than rethrowing. I think it's confusing when an
exception from one thread is rethrown in another thread since the stack traces
don't match up. It was my bad though not checking that I hadn't broken version
negotiation beofre committing on QPID-1339, so I'll reassign to myself and play
with some other options. I have a vague idea that cloning the original
exception and then reinitializing the stack trace and initial cause might
result in generic way to create an exception that preserves the stack traces
from both threads, but doesn't require the handling code to do a DFS through a
linked chain of exceptions in order to get to the detail associated with the
originating chain of exceptions.
> low level refactoring for replay breaks protocol negotiation
> ------------------------------------------------------------
>
> Key: QPID-1344
> URL: https://issues.apache.org/jira/browse/QPID-1344
> Project: Qpid
> Issue Type: Bug
> Components: Java Common
> Affects Versions: M4
> Reporter: Martin Ritchie
> Assignee: Rafael H. Schloming
> Fix For: M4
>
>
> Summary:
> The change in QPID-1339 is causing Protocol negotiation to fail.
> Problem
> The initial problem is that the check for protocol 0-10 in ClientDelegate was
> incorrect preventing the ProtocolVersionException from being propogated to
> the connection.
> The new Connection class also wraps the ProtocolVersionException so
> AMQConnection is unable to work out the new protocol to try so it continues
> to try 0-10 until failover is exhausted.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
